Zone II • Versions EN3
Abstract: ENVI 5.2 and ArcGIS 10.1 were used to process GF-2 image, Sentinel-2 image and Landsat TM/OLI image data covering Sanya Coral Reef National Nature Reserve. Assisted by field research data, we used visual interpretation and threshold segmentation to summarize the rules of coral reef distribution,and to extract the spatial distributional data of shallow sea coral reefs in the Danzhou Bay and Sanya Coral Reef National Nature Reserve. The accuracy of the results was verified by field spectrum information collected. Finally, we obtained the spatial distribution data of shallow sea coral reefs in phase 51. This data set can be used for the analysis of the temporal and spatial changes, eco-environmental changes of coral reefs, among others. With the support time series, it clearly reflects the distributional changes of coral reefs in Sanya National Nature Reserve.
Keywords: coral reef; remote sensing monitoring; visual interpretation; threshold segmentation; information extraction
|Title||Distributional data of shallow sea coral Reef in Danzhou Bay and Sanya Coral Reef National Nature Reserve from 1987 to 2018|
|Data corresponding author||Zhu Lanwei(firstname.lastname@example.org)|
|Data authors||Huo Yanhui, Zhu Lanwei, Zhang Shaoyu, Yang Xu, Tang Shilin|
|Geographical scope||This data set has a geographic scope of 18°10'30"N - 18°15'30N", 109°20'50"E- 109°40'30"E and 19°33'00"N - 19°53'00N", 108°57'00"E- 109°16'00"E. Located in the south and the west coast of Hainan Province respectively.|
|Spatial resolution||4 m, 10 m, 30 m|
|Data volume||14.1 MB|
|Data service system||<http://www.sciencedb.cn/dataSet/handle/687>|
|Sources of funding||Major Science and Technology Program of Hainan Province (ZDKJ2016021); Genneral Programs of Natural Science Foundation of Hainan Province under Grant No.sy17zm01132; Open Fund of Guangdong Key Laboratory of Ocean Remote Sensing under Grant No. 2017B030301005-LORS1904; Major Program for Big Data Development of the National Development and Reform Commission (2016-999999-65-01-000696-01).|
|Dataset composition||The data set is mainly including Hainan Danzhou Bay and Sanya Coral Reef National Nature Reserve, three different data sources of shallow sea coral reefs spatial data products, these data is saved as a compressed file (Data set of shallow coral reef distribution in Danzhou bay and Sanya coral reef national nature reserve between 1987 to 2018. Rar), each data exist alone named after years of space distribution of coral reef folder.|
Coral reefs, known as "oases in the blue desert", are a very unique ecosystem in the ocean. Species living in coral reefs are extremely abundant. Coral reef is a unique marine ecosystem with rich biodiversity, as well as an important fishing ground and marine tourism resources. Coral reefs have high ecological and economic values. Coral reefs account for about 0.25% of the world's total ocean area, but they feed more than a quarter of the world's marine fish. Coral reef ecosystem has amazing biodiversity and high primary productivity. It is very sensitive to sea water temperature, acidity and pollutants. It can be used as a main biological indicator of marine environment. It can also play a role in regulating and optimizing the marine environment. It is of great significance to timely grasp the spatial distribution of coral reefs for monitoring the marine ecological environment.
Based on Multi-temporal Remote Sensing data, this paper uses visual interpretation and threshold segmentation method to obtain the data products of coral reef spatial distribution in Danzhou Bay and Sanya National Nature Reserve from 1987 to 2018, and provides data sharing services. Relevant research results, as part of the coastal ecological environment monitoring in Hainan Province, reflect the spatial distributional characteristics of coral reefs under time series, and provide scientific data support for the protection and management of coral reefs in protected areas.
The data set is based on the GF-2 multispectral image of Sanya in Hainan Province with a resolution of 4 meters in 2017 provided by the Hainan High Fraction Data and Application Center, the 10-meter resolution Stinel-2 multispectral image of Hainan from 2016 to 2018 provided by the European Space Agency (http://scihub.copernicus.eu/) and the 30-meter resolution of Hainan from 1987 to 2018 provided by the United States Geological Survey (http://glovis.usgs.gov/). The Landsat TM/OLI multi-spectral remote sensing image is the data source. A total of 44 images with no or few clouds are obtained. Table 1 shows the serial number and imaging time information of the remote sensing satellite image used in this time.
|Number||Data||Satellite||Sensor||Image Sequence Number|
2.2.1 Visual interpretation
Image interpretation refers to the basic process of obtaining information from images. Visual interpretation is a kind of remote sensing image interpretation. To put it another way, visual interpretation is the reverse process of remote sensing imaging. Visual interpretation is the process of recognizing targets from remote sensing images by means of image interpretation marks, extracting the information of distribution, structure and function of targets qualitatively and quantitatively, and expressing them on the geographic map.
Image interpretation marks, also known as interpretation elements, are image features that can directly reflect and discriminate the information of topographic features in remote sensing images, including shape, size, color and tone, shadow, position, structure, texture, pattern, layout and three-dimensional appearance, etc..
Combining the characteristics of optical color, geographical location, shape and generating conditions of shore reefs on remote sensing images, a comprehensive and omni-directional interpretation mark is formed. With the help of remote sensing enhancement methods (such as image fusion, image stretching, principal component transformation, etc.), Landsat TM/OLI images are displayed in ArcGIS 10.1 software according to the optimal band combination and image enhancement mode, combined with field survey. Points to establish interpretation markers for shallow coral reefs is shown in Table 2.
The visual interpretation data set acquires the spatial distribution information of shallow sea coral reefs according to the processing flow shown in Figure 1, loads the image in ENVI, first carries on the radiation calibration, transforms the DN value into the radiation brightness data, obtains the radiation brightness image; then carries on the FLAASH atmospheric correction, transforms the radiation brightness value into the surface reflectance, obtains the surface reflectance image; and then carries on the tailoring according to the protected area. Secondly, the normalized difference water body index (MNDWI) is calculated, the histogram threshold is segmented, and then the land-water mask is carried out to remove the land mask. In ArcGIS 10.1, the images passing through the land mask are opened according to the combination of red, green and blue bands. Based on the visual interpretation signs determined in this section, the spatial distribution information of shallow sea coral reefs is obtained by digital vectorization method. Finally, the spatial distribution results of shallow-sea coral reefs in the 17-stage Danzhou Bay and the 19-stage Shallow-sea coral reefs in the Sanya Coral Reef National Nature Reserve were obtained.
|Interpretation mark||Name||Image characteristics||Mark detail|
|Location||Crack||Coastal sediment is too much to grow coral reefs. There is a channel gap between the reef and the coast with a depth of about 0.2-1.5 meters.|
|Color||Beach||In the RGB true color composite image, the beach color is blue, and the coral reef color is yellow-green. Note the distinction between beach and coral reef colors, which are easily confused.|
|Shape||Striped coral reefs||Coastal reefs are reefs growing along the coast of continents or islands, generally showing strip or band distribution.|
|Location||River estuaries||Reef-building corals grow in seawater with salinity of 27 to 40 and the optimum salinity range is 34 to 36. Where freshwater enters the sea, the salinity will be diluted and reduced, which is not suitable for coral reef formation.|
|Depth of water||Depth of water||It is generally believed that the range of water depth for reef-building corals is 0-50 m, and the optimum water depth is less than 20 m.|
2.2.2 Automatic Extraction of Threshold Segmentation
Firstly, the original remote sensing image is radiometric calibrated, and DN value is converted into radiance data to obtain radiance image; after atmospheric correction of the calibrated remote sensing image, the radiance value is converted into surface reflectance, and the surface reflectance image is obtained; then, the coastline data and normalized difference water index (MNDWI) are used for threshold segmentation to mask land information. The spatial distribution information of coral reefs in shallow sea was obtained by histogram threshold segmentation using blue-green band ratio and normalized vegetation index (NDVI). The extraction process is shown in Figure 2.
The spatial distribution data of shallow sea coral reefs in Danzhou Bay and Sanya National Nature Reserve were obtained by threshold segmentation method (based on Landsat 8 images from 2013 to 2018).
Based on Sentinel-2 image from 2016 to 2018, the spatial distribution data of shallow sea coral reefs in Sanya Coral Reef National Nature Reserve were obtained by threshold segmentation method.
3.1 Data composition
This data set includes 51 periods of shallow sea coral reef spatial distribution data of Danzhou Bay and Sanya Coral Reef National Nature Reserve from 1987 to 2018. These data are stored in a compressed file ("Shallow sea coral reef distribution data set of Danzhou Bay and Sanya Coral Reef National Nature Reserve from 1987 to 2018.rar"). Each period of data is stored separately in a named folder for each year, totaling the total number. The volume is 14.1 MB. The corresponding SHP vector data files are stored in each folder. All data are in UTM-WGS84 coordinate system consistent with remote sensing images.
3.2 Data sample
Spatial distribution data of shallow coral reefs in Danzhou Bay are shown in Fig. 3-5. Spatial distribution data of shallow sea coral reefs in Sanya Coral Reef National Nature Reserve are shown in Fig. 6-11.
4.1 Accuracy Verification Method and Technical Route
The data set validates the accuracy by validating the fitting degree between the spectral reflectance curve of coral reefs on image and the measured spectral reflectance curve of coral reefs.
Firstly, spectral reflectance curves of coral reefs are extracted from images. Using ENVI software, the average spectral reflectance curve of coral reef patches was calculated and the average spectral reflectance was extracted.
Secondly, the spectral reflectance curves of coral reefs are acquired and generated by the portable ground object spectrometer (FieldSpec 3 produced by ASD Company, USA). The spectrum of coral reef was collected by underwater measurement method, and the spectrum was processed by viewspec Pro software. The spectral reflectance curve of coral reef was obtained. The spectral reflectance values corresponding to the central wavelength of the image are extracted.
Finally, the corrcoef (x, y) function of MATLAB is used to calculate the correlation coefficient R between the average spectral reflectance of coral reefs and the spectral reflectance of corresponding wavelengths. The value of R in [-1,1], 0 means irrelevant; 0-1 means positive correlation, the larger the value, the more relevant; 1-0 means negative correlation, the smaller the value, the more relevant.
4.2 Accuracy verification results
Taking the Landsat 8 image of 2018 as an example, five patches (Xidaimao Island, Luhuitou Bay, Xiaodong Sea, Linqiangshi Island and Eman Port) with concentrated coral reefs were selected from the study area, and their average spectral reflectance was calculated. The calculated correlation coefficient R is shown in Table 3.
As can be seen from Table 3, the correlation coefficients R of the above five patches are all above 0.8, with the highest reliability of Eman Port being 96%, and the lowest reliability of Xiaodonghai being 86%. Overall, the reliability of coral reef extraction target can reach more than 80%.
From 1987 to 2018, the spatial distribution data of shallow-sea coral reefs in Danzhou Bay and Sanya Coral Reef National Nature Reserve Phase 51 is SHP format. ArcGIS, ENVI and other related data processing software can be used to edit, query and follow-up analysis of the data set. This data set can provide basic data support for coral reef protection and regulation.
Thank the project team for the support and cooperation of relevant industry departments and units in Hainan Province in the field observation process, and express our heartfelt thanks. I would like to thank researchers Zhang Li and Liao Jingjuan for their methodological suggestions in validating the spatial distribution information of coral reefs. I would like to thank Guo Lianjie, Bi Jingpeng, Song Qixi, Zhang Yunfei and others for their work and contributions in image downloading, field investigation and technical route discussion.
Pan Yanli, Tang Danling. Overview of coral reef albinism by satellite remote sensing [J]. Journal of Ecology, 2009, 29 (09): 5076-5080.
Chen Jianyu, Mao Zhihua, Zhang Huaguo, et al. Remote Sensing Capability Analysis of Dongsha Atoll Reef Based on SPOT5 Data[J]. Journal of Oceanography (Chinese Edition) , 2007 (03): 51-57.
Zou Yarong, Liang Chao, Zhu Haitian. Experimental study on monitoring coral reef development based on QuickBird image [J]. Journal of Oceanography (Chinese edition), 2012, 34 (02): 57-62.
Chen Guohua, Huang Liangmin, Wang Hankui, et al. Progress in primary productivity of coral reef ecosystem [J]. Journal of Ecology, 2004, 24 (12): 2863-2869.
Yu Kefu. Coral reefs in the South China Sea and their records and responses to Holocene environmental changes [J]. Chinese Science: Geosciences, 2012 (8).
Chen Jianping, Miaofang. Research status and development trend of remote sensing image interpretation [J]. Land and Resources Remote Sensing, 2004, 16 (2).
Pu Jingjuan. Principles and methods of visual interpretation of remote sensing images [M]. Beijing: China Science and Technology Press, 1992.
Jiang Fang, Zhang Guoyong. Several issues deserving attention in visual interpretation of remote sensing images [J]. Journal of Changchun Institute of Engineering: Natural Science Edition, 2002, 3 (3): 49-50.
Hu Leiqiu, Liu Yalan, Ren Yuhuan, et al. Study on Information Extraction Method of Nansha Coral Reef from SPOT5 Multispectral Image[J]. Remote Sensing Technology and Applications, 2010, 25 (04): 493-501.
Huo Yanhui, Zhu Lanwei, Zhang Shaoyu, Yang Xu, Tang Shilin. Distributional data of shallow sea coral reef in Danzhou Bay Sanya Coral Reef National Nature Reserve from 1987 to 2018 [DB/OL]. Science Data Bank, 2018. (2018-11-20). DOI: 10.11922/sciencedb.687.
How to cite this article
Huo Yanhui, Zhu Lanwei, Zhang Shaoyu, Yang Xu, Tang Shilin. Distributional data of shallow sea coral reef in Danzhou Bay Sanya Coral Reef National Nature Reserve from 1987 to 2018 [J/OL]. Chinese Scientific Data, 2018. (2018-12-07). DOI: 10.11922/csdata.2018.0080.zh.